Abstract:Negative sampling methods are vital in implicit recommendation models as they allow us to obtain negative instances from massive unlabeled data. Most existing approaches focus on sampling hard negative samples in various ways. These studies are orthogonal to the recommendation model and implicit datasets. However, such an idea contradicts the common belief in AutoML that the model and dataset should be matched. Empirical experiments suggest that the best-performing negative sampler depends on the implicit dataset and the specific recommendation model. Hence, we propose a hypothesis that the negative sampler should align with the capacity of the recommendation models as well as the statistics of the datasets to achieve optimal performance. A mismatch between these three would result in sub-optimal outcomes. An intuitive idea to address the mismatch problem is to exhaustively select the best-performing negative sampler given the model and dataset. However, such an approach is computationally expensive and time-consuming, leaving the problem unsolved. In this work, we propose the AutoSample framework that adaptively selects the best-performing negative sampler among candidates. Specifically, we propose a loss-to-instance approximation to transform the negative sampler search task into the learning task over a weighted sum, enabling end-to-end training of the model. We also designed an adaptive search algorithm to extensively and efficiently explore the search space. A specific initialization approach is also obtained to better utilize the obtained model parameters during the search stage, which is similar to curriculum learning and leads to better performance and less computation resource consumption. We evaluate the proposed framework on four benchmarks over three models. Extensive experiments demonstrate the effectiveness and efficiency of our proposed framework.

A Gain-Tuning Dynamic Negative Sampler for Recommendation

MixGCF

Dynamic negative sampling for recommendation with feature matching

Enhancing Recommender Systems: A Strategy to Mitigate False Negative Impact

Towards Automated Negative Sampling in Implicit Recommendation

Generating Negative Samples for Sequential Recommendation

Reinforced Negative Sampling for Recommendation with Exposure Data

Revisiting Negative Sampling Vs. Non-sampling in Implicit Recommendation

Optimizing Top-N Collaborative Filtering Via Dynamic Negative Item Sampling

Reinforcement negative sampling recommendation based on collaborative knowledge graph

On User-side Fairness in Negative Sampling for Recommender Systems

Bayesian Negative Sampling for Recommendation

Fairly Adaptive Negative Sampling for Recommendations.

ENSG: Enhancing Negative Sampling in Graph Convolutional Networks for Recommendation Systems

Mixed Negative Sampling for Learning Two-tower Neural Networks in Recommendations.

GANRec: A Negative Sampling Model with Generative Adversarial Network for Recommendation

Personalized Negative Reservoir for Incremental Learning in Recommender Systems

On the Theories Behind Hard Negative Sampling for Recommendation

ABNS: Association-based negative sampling for collaborative filtering

Batch-Mix Negative Sampling for Learning Recommendation Retrievers

Negative Sampling in Recommendation: A Survey and Future Directions